NONDESTRUCTIVE EVALUATION OF PLEXIGLAS MATERIALS USING R. Montanini

XIX IMEKO World Congress
Fundamental and Applied Metrology
September 6−11, 2009, Lisbon, Portugal
NONDESTRUCTIVE EVALUATION OF PLEXIGLAS MATERIALS USING
LOCK-IN AND PULSE PHASE INFRARED THERMOGRAPHY
R. Montanini 1, S. Aliquò 1
1
Faculty of Engineering, University of Messina, Messina, Italy, [email protected]
Abstract − Lock-in and pulse phase infrared
thermography measurement techniques have been exploited
for quantitative assessment of subsurface defects in a
reference specimen made of Plexiglas. Radiometric thermal
images were post-processed using a contrast approach in the
frequency domain, allowing defect depth location to be
resolved with a relative combined standard uncertainty of
about 7% for thicknesses above 3 mm. Conversely,
significant radial heat diffusion next to the boundary of the
discontinuity made accurate sizing of deeper subsurface
defects more difficult. The obtained results demonstrate the
potentiality of active thermography as a fast, powerful
contactless NDT measurement tool.
Keywords: Temperature measurements, radiometric
infrared thermography, non destructive testing (NDT)
1. INTRODUCTION
Total quality is now an established concept for mass
products such as cars, consumer electronics, and personal
computers. In many fields, primarily aerospace and military,
it has been the rule for years, for security reasons. A major
effort to reach this quality concept is to implement
inspection tasks along the production line through effective
nondestructive testing (NDT) methods to be used for either
acceptance/rejection of parts, or for inspection of end
products in service [1].
In recent years, remarkable progresses in high sensitive
infrared image detectors and much effort from researchers in
leading university laboratories have brought about fast
growth of infrared radiometric measurement techniques.
Nowadays, active thermography [2-7] is recognized as one
of the most powerful NDT tool to detect flaws and defects in
different kinds of materials, such as metals, composites, and
polymers. The active approach involves using an excitation
source to induce thermal contrast into the material and an IR
camera to measure the stationary or transient response.
In particular, lock-in thermography [8-10] makes use of
modulated optical stimulation to provide information about
the thermophysical properties of the material as well as to
identify subsurface defects in a quick and contactless way.
Lock-in thermography has been proposed to detect areas of
disbond in coatings [2], delaminations, impact damage and
inclusions of spurious materials in composite materials
[2,4,6], flat-bottom hole defects in steel [11-12],
delaminations of veneered wood [13], to visualize fibre
ISBN 978-963-88410-0-1 © 2009 IMEKO
orientation in composites [14], to identify detachments or
cracks in frescoes [15-17]. One limitation of lock-in
thermography lies on the available range of frequencies for
the heat flux modulation, which may be not sufficiently low
to detect deeper defects in materials of very low thermal
diffusivity and/or large thickness.
Another approach is pulse thermography, which can be
obtained by stimulating the part with one or more pulsed
heat sources and monitoring its surface temperature
evolution during the transient heating (cooling) phase
[3,6,11,15,16]. The visibility of defects depends on several
factors, which include material characteristics (i.e. thermal
contrast), atmospheric conditions and instrument sensitivity.
Measurements performed by means of the pulsed
thermographic method are affected by local variation of the
emissivity coefficient and by non-uniform heating of the
surface, that can mask the defect visibility. The emissivity
problem may be overcome by painting the surface, but this
could be a solution only for parts where this surface finish is
tolerable. The uniformity of surface heating may be
improved by using the lateral heating technique described
by Grinzato et al [17].
A measurement technique which combines the
advantages of both lock-in and pulse thermography without
sharing their drawbacks is pulse phase termography, which
has been first proposed by Maldague and Marinetti [18-19].
The specimen is pulse heated as in pulsed thermography and
the mix of frequencies of the thermal waves launched into
the specimen is unscrambled by computing the Fourier
transform of the temperature evolution over the field of
view; the phase, or magnitude, image can be presented as in
modulated lock-in termography. The fact that pulse phase
termography sorts available information coherently in term
of frequencies brings interesting features with respect to the
more traditional contrast approach used in pulse
thermography.
In this paper both lock-in and pulse phase infrared
thermography measurement techniques have been
implemented for quantitative assessment of subsurface
defects in a reference specimen made of Plexiglas. By
proper post-processing of phase thermal images, size and
depth locations of flat-bottom hole defects were determined
and the measurement uncertainty assessed. Merits and
limitations of the proposed approach, as well as metrological
aspects related to possible interference inputs, are discussed
thoroughly.
1524
2. EXPERIMEN
E
NTAL METH
HODS
All the expeerimental testss were perform
med using a 14-bit
1
diggital output infrared cam
mera (Cedip Infrared Systtems,
Crroissy Beaubbourg, Francce), equippeed with a high
ressolution (320×240 pixel) InSb
I
focal plaane array and a 25
mm
m optical lenss. The IR cam
mera operates in the MWIR
R (3-5
µm
m) wavelength spectral rannge and has a noise equivvalent
tem
mperature diff
fference (NET
TD) of 20 mK at 25 °C. Thee data
accquisition systtem is controlled by a persoonal computerr and
alllows the integgration time too be varied inn the range 10 μs –
5 ms,
m with fram
me rates up to 80
8 Hz at full resolution.
r
Synchronizaation of extternal analoggue signals were
acccomplished by
b the lock-iin module inntegrated intoo the
inffrared cameraa.
Images acqquisition, cam
mera configuuration and nonunniformity corrrection (NUC) calculationss were carriedd out
ussing the DispplayIMG ver.2.6 softwaree (Edevis Gm
mbH,
Sttuttgart, Germ
many). Image post-processinng was perforrmed
wiith the Labvieew™ ver. 8.5 software (Nattional Instrum
ments,
Auustin, Texas).
mera signals were
w
synchroniized using a lock-in
l
ampliffier
cam
(actu
ually integrateed in the cameera).
To
T heat the saample, a sinussoidal modulated optical waave
is used, thus prooducing a therrmal wave which
w
propagaates
throu
ugh the thickkness of the object. As th
he thermal waave
enco
ounters a disccontinuity it uundergoes refl
flections becauuse
of the locally chhanged heat ppropagation parameters.
p
T
The
refleected wave innterferes with the surface wave
w
giving rise
r
to a stationary oscillating ttemperature field, which is
remo
otely measureed through its thermal infraared emission by
the IR camera. Amplitude and phase images of the
onstructed therrmal wave weere computed
d in real-time for
reco
each
h heat-generatting frequencyy using the Fourier transfoorm
algo
orithm. Thesee two quantitties are used
d to present the
relev
vant information about ssubsurface discontinuities. A
phottograph of a typical expeerimental set-up is shown in
Fig.3
3.
IR-camera
Power amplifier
Function generaator
nce specimen preparation
p
2.1. Referen
A
referennce PMMA
A
(PolyMetthylMethAcryylate)
sppecimen, with known flat-bbottom hole defects
d
of diffferent
deepth, has beenn prepared as shown in Figg.1. The speccimen
haas a circular shape
s
and preesents sixteenn 10 mm diam
meter
hooles, whose deepth from thee front side suurface ranges from
0,66 mm (top right corner) too 3,6 mm (boottom left corrner).
Thhe specimen front
f
surface was painted with
w opaque black
b
paaint to uniform
m emissivity annd reduce refllections.
halogen lamps
Frame grabbeer
optical wa
ave
FFT analyzer
Lock‐in amplifiier
Computer
defect
reflected
thermal wave
thermal input
Fig.
F 2. Lock-inn infrared therm
mography measu
uring principle.
a
a)
b)
Fig. 1. PMMA
A reference speecimen with flatt-bottom holes of
diifferent depth: a) rear side of the
t specimen, b)
b drawing show
wing
d
and relative positioon of defects.
nominal dimensions
2.2. Lock-in
n termographyy
The measuuring principlle used to carry
c
out lock-in
meeasurements is
i schematizedd in Fig. 2. The
T IR cameraa was
cooupled to a heeat source (4×
×1 kW halogeen lamps) whiich is
drriven by a power
p
ampliffier and a fuunction generrator.
Haalogen lamps were positionned so as to produce
p
a uniiform
disstribution off heat across the illuminnated side off the
sppecimen. The distance of the
t IR-cameraa from the saample
waas approximaately 50 cm in order to record the entire
e
suurface of the sample. Thee heat source and the infr
frared
Fig. 3. Experrimental set-up for lock-in measurements.
2.3.
2 Pulse phaase thermograaphy
The
T measurinng principle uused to carry out pulse phaase
meaasurements is schematized in Fig. 4. Insstead of makiing
use of a sinusoidaal modulation, the specimen
n is pulse heaated
usin
ng either haloggen lamps (loong pulse > 500
5 ms) or hiigh
pow
wer xenon tubee ring flashes (short pulse < 1 ms). Hennce,
therm
mal waves off different freqquencies are laaunched into the
1525
specimen simultaneously and the resulting temperature field
of its surface is measured in the transient regime.
Extraction of the various frequencies was performed by
acquiring sequences of images and extracting the temporal
decay of each pixel in the field of view. Hence, the discrete
one-dimensional Fourier transform (DFT) was applied on
each pixel of the thermogram sequence to compute the real
and imaginary parts and, finally, amplitude and phase
images were calculated.
IR-camera
Power amplifier
Pulse generator
flash lamps
Frame grabber
FFT analyzer
optical pulse
reflected thermal pulse
defect
thermal pulse
Computer
thermal input
Fig. 4. Pulsed infrared thermography measuring principle.
3. RESULTS AND DISCUSSION
Lock-in amplitude and phase images of the reconstructed
thermal wave on the reference specimen surface are reported
in Fig. 5 for different modulation frequencies.
As it can be observed, the information given by the
phase image is actually more effective, since it is relatively
independent of local optical (e.g. non-uniform heating) and
infrared (e.g. variability in surface emissivity) surface
features. As a consequence, only signal amplitude is affected
by the specimen topography while phase is not, except for
the level of phase noise which of course increases in parts
where less light is absorbed per unit area.
At 2,250 Hz, the relatively high modulation frequency
limits the analysis to a close to the surface region, where
black coating disuniformity can be appreciated in the phase
image. By reducing the lock-in frequency, deeper defects are
progressively revealed.
In the mid-high range (0,100 Hz < f < 0,025 Hz) the first
two rows are clearly visible: in the phase images, defects
appear brighter than the surrounding background and their
boundaries are well defined.
The third row starts to come out for modulation
frequencies in the mid-low range (0,015 Hz < f < 0,008 Hz).
At this stage, the sharpness of the shallower defects in the
first row becomes poor.
At 4 mHz, also the deepest flat-bottom hole is detected,
although the contrast with the background is remarkably
lowered due to a noteworthy reduction in the phase shift.
Moreover, at very low frequency the boundary of the defects
in the thermal image seems to be “out of focus”, because of
significant radial heat diffusion that takes place in the
specimen plane.
This frequency-dependent behaviour is a direct
consequence of the inverse relationship existing between the
thermal diffusion length µ and the modulation frequency f :
μ=
2k
ρ ω Cp
=
α
π f
(1)
being
2 , k the thermal conductivity of the
material, ρ the density, Cp the specific heat at constant
pressure and α the thermal diffusivity. For the amplitude
image, the allowable depth range is given by (1) while, for
the phase image, the maximum depth that can be inspected
actually corresponds to 1,8µ [9]. Hence, usually tests should
start at a quite high frequency value to investigate the
surface layer; then the frequency must be decreased to
investigate a deeper layer, after which the frequency must be
further reduced. This procedure must go on until the entire
thickness of the object is investigated or the minimum
available modulation frequency is reached.
Similar results (Fig. 5, last column) were obtained by
unscrambling the frequency content of the infrared images
sequence measured after thermal pulse stimulation (pulse
phase approach). With respect to the lock-in technique, this
method can be much more fast, since it might need just one
measure to analyze the whole frequency range of interest,
whereas the lock-in approach requires multiple iterative
tests. Phase contrast however is poorer, hiding the detection
of deeper flaws. One further drawback of the pulse phase
approach is that a definite temperature difference between
two successive images of the sequence must exist to clearly
discern defects. To display discontinuities located more in
depth a higher surface heating is usually needed.
For quantitative non-destructive evaluation of subsurface
defects the thermal diffusivity of the material should be
known, as pointed out in (1). Unfortunately, the exact value
of this parameter is not always available in the literature,
because it strongly depends on the actual material
composition.
To overcome this problem, a fast method which makes
use of a high speed IR camera has been developed. The
proposed approach is based on a transmission scheme with
thermal pulse stimulation and on the well known Parker’s
law [20]:
α = 1,37
l2
π 2 t 0, 5
(2)
where l is the material thickness and t0,5 the half
temperature rise time, that is the time needed to reach one
half of the maximum temperature increment over the sample
surface opposite to the heated one. A high power flash was
used to generate a sudden energy pulse (to limit heat
exchange with the surroundings), so that test conditions
were not too dissimilar from ideal adiabatic ones.
The IR camera recorded the thermal evolution, allowing
for the computation of the half temperature rise time.
The accuracy of the method can be further improved by
using specific non-linear interpolation models which take
into account the actual heat exchange conditions [21].
1526
AMPLITUDE LI
PHASE LI
A
AMPLITUDE
LI
L
PHASE LI
PHASE PP
f = 2,250 Hz
f = 0,100 Hzz
f = 0,100 Hz
f = 0,080 Hz
f = 0,040 Hzz
f = 0,040 Hz
f = 0,025 Hz
f = 0,015 Hzz
f = 0,015 Hz
f = 0,008 Hz
f = 0,004 Hzz
f = 0,004 Hz
Temperature increment / °C
Temperature increment / C
Fig. 5. Lock--in IR measurem
ments (LI): ampplitude (left) annd phase (right) images of the reconstructed
r
thhermal wave on
n the reference
sppecimen surfacee obtained at diffferent modulattion frequenciess. Pulse phase IR measurements (PP): phase iimages obtained
d by unscrambling
the frequenccy content of thhe infrared imagges sequence measured
m
after th
hermal pulse stiimulation usingg discrete Fourieer transform.
1,4
1,3
1,2
1,1
1,0
0,9
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0,0
0
t1/210
20
30
40
0
50
60
70
80
90
100
Time / s
Fig.
F 6 shows the
t temperatuure evolution measured
m
oveer a
small area of a defect-free
d
sam
mple. It can be
b noted the low
l
dispersion of the experimental
e
data and the negative
n
slopee of
the curve
c
after the maximum temperature haas been reachhed,
due to the coolinng effect. Sinnce the meassurement of the
therm
mal transient used for the determination
n of the therm
mal
diffu
usivity is perfformed by thee IR camera, local values can
c
be averaged,
a
thus improving thhe S/N ratio.
According
A
to (1) the actuall depth of thee defect depennds
upon
n the modulattion frequencyy at which th
he defect itselff is
first detected.
To
T determinne unambiguuously this frequency, the
norm
malized contraast was first deefined as follo
ows:
C (t ) =
Fig. 6. Measurem
ment of the maaterial thermal diffusivity
d
by means
m
of a fast IR trransient flash method:
m
normaliized temperaturre
increments vs.
v heating timee.
1527
S def (t ) − S def (t0 )
S b ( t m ) − S b (t 0 )
(3)
0,20
Def #1
0,18
Def #2
Def #3
0,16
Def #4
0,14
Def #5
0,12
Def #6
Def #7
0,10
Def #8
0,08
Def #9
0,06
Def #10
Def #11
0,04
for estimated defect depth values were instead computed as
combined standard uncertainty considering the type A
contributions of fTR and α.
Relative uncertainty of depths estimation ranges from
5,4% (defect #6) to 7,2% (defect #14). From Fig. 8 it can be
observed that, as the two deepest defects are concerned, the
uncertainty bands do not overlap. It can be concluded that
3,2 mm thickness roughly corresponds to the maximum
detection limit for the current measurement setup.
As far as defects size assessment is concerned, obviously
the best accuracy can be achieved when the contrast
between defected and undefected areas is maximized.
Hence, starting from the normalized contrast vs.
modulation frequency plots, the frequency fmax (for which it
results C = Cmax) was first estimated for each region of
interest (Fig. 9).
Def #12
0,02
Def #13
0,00
Def #14
0,001
0,01
0,1
Modulation frequency / Hz
1
0,20
Def #16
Fig. 7. Family of normalized contrast curves computed for each
area of interest according to (3) vs. modulation frequency.
To determine the frequency at which each defect had
been first detected (fTR), a threshold level CTR was defined,
so that for C ≥ CTR the discontinuity was assumed
detectable, while for C < CTR it was not. The threshold value
was determined by an iterative procedure.
Finally, the subsurface defect depth was calculated as
p = 1,8μ = 1,8
α
π fTR
3,50
11
3,00
3
3
6
7
10
11
9
13
2,50
22
5
9
5
13
6
10
14
14
7
11
15
15
4
8
4
8
12
16
12
16
Def #3
Def #4
0,14
Def #5
0,12
Def #6
Def #7
0,10
Def #8
0,08
Def #9
0,06
Def #10
Def #11
Def #12
0,02
Def #13
Def #14
0,00
0,01 f(C3_max) f(C4_max) 0,1
0,001
Modulation frequency / Hz
1
Def #15
Def #16
Fig. 9. Determination of the maximum normalized contrast
modulation frequency for each region of interest.
Then, the nearer available (i.e., such that f ≅ fX_max) lockin phase image was selected. The image file was hence postprocessed through different steps (filtering, image
calibration, overlaying of predefined geometrical objects,
measure of the defect area) by means of a dedicated digital
image processing software written in Labview™.
Results are reported in Fig. 10, where they are compared
with the nominal radius value.
2,00
8
1,50
1,00
0,50
0,00
DEF DEF DEF DEF DEF DEF DEF DEF DEF DEF DEF DEF DEF DEF DEF DEF #4 #3 #2 #1 #8 #7 #6 #5 #12 #11 #10 #9 #16 #15 #14 #13
Nominal depth 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 2,8 3,0 3,2 3,4 3,6
Estimated depth 0,7 0,9 1,0 1,1 1,3 1,6 2,0 2,0 2,4 2,6 2,5 2,6 2,8 3,1 3,1 3,2
Fig. 8. Quantitative nondestructive evaluation of subsurface
defects: comparison of estimated and nominal defects depth.
Subsurface defect radius / mm
Subsurface defect depth / mm
4,00
Def #2
C3_max
0,16
0,04
(4)
Fig. 8 reports the comparison between estimated and
nominal defects depth.
Def #1
C4_max
0,18
Def #15
Normalized contrast
Normalized contrast
where Sdef and Sb are the phase signal levels of defected
and background areas, respectively, measured at time t and
t0, and tm is the time at which the heat absorption is
maximum.
Then, for each area of interest, C values at time t were
plotted as a function of the modulation frequency (Fig. 7).
7
Nominal radius: 5 mm
6
5
4
3
2
1
11
22
3
4
55
66
77
88
10
10
11
11
1212
14
15
99
13
13
14
3
4
15 1616
0
Uncertainty bands for nominal depth values were
calculated as type B (UNI CEI ENV 13005) by assuming a
rectangular distribution of width 2R, where R = 0,05 mm is
the resolution of the calliper used to measure the depth of
the reference specimen flat-bottom holes. Uncertainty bands
1528
Fig. 10. Quantitative nondestructive evaluation of subsurface
defects: comparison of estimated and nominal defects size.
Uncertainty bands were calculated as twice standard
deviations (95% confidence value) on three measurements
carried out with slightly different geometrical circles in
order to take into account data variability induced by
unsharpened defect boundaries.
Relative uncertainty ranges from 4,5% for the shallower
defects in the first row to about 88% for the deepest one
(defect #13). The relative high uncertainty found, in
particular for the deeper defects (located in the lower two
rows), can be attributed to the limited number of
experimental data (i.e., modulation frequency steps) and to
significant radial heat diffusion, that made measurement of
the actual hole boundary quite difficult.
It is worth mentioning that similar results can be found
by post-processing the thermograms obtained using the
pulse phase approach.
[4]
[5]
[6]
[7]
[8]
[9]
4. CONCLUSIONS
[10]
Active thermography measurement techniques have been
exploited for quantitative assessment of subsurface defects
in a reference specimen made of Plexiglas. By proper postprocessing of phase thermal images recorded at different
frequencies and direct measurement of material’s thermal
diffusivity, a fast, contactless and effective NDT
methodology has been demonstrated.
Lock-in thermography provides quantitative information
about size, depth and thermal resistance of defects and, as
phase images are used, relatively insensitivity to nonuniform heating and local variation of the emissivity
coefficient. The main limitation of this technique lies in the
minimum frequency for the heat flux modulation which may
be not sufficiently low to detect deeper defects.
Pulse phase thermography is even more fast than lock-in.
However, there are also some limitations: the main one is
that to display deeper defects a higher surface heating is
usually required, which may damage plastic materials.
Future work directions will addressed the evaluation of
different types of artificial defects as well as natural ones.
The authors would like to thank Prof. G. Busse and his
research group at the IKP-ZfP Insitute of the University of
Stuttgart (D) for their kindly support and precious advice.
REFERENCES
[2]
[3]
[12]
[13]
[14]
[15]
[16]
[17]
ACKNOWLEDGMENTS
[1]
[11]
[18]
[19]
[20]
X. Maldague, Theory and practice of infrared technology for
nondestructive testing, John Wiley Interscience, New York,
2001.
D. Wu, G. Busse, “Lock-in thermography for nondestructive
evaluation of materials”, Rev. Gén. Therm., vol. 37, pp. 693–
703, 1998.
C. Meola, G.M. Carlomagno, Recent advances in the use of
infrared thermography”, Meas. Sci. Technol., vol. 15, pp.
R27–R58, 2004.
[21]
1529
C. Meola, G.M. Carlomagno, A. Squillace, G. Giorleo,
“Non-destructive control of industrial materials by means of
lock-in thermography”, Meas. Sci. Technol., vol. 13, pp.
1583–1590, 2002.
C. Meola, G.M. Carlomagno, L. Giorleo, “The use of
infrared thermography for materials characterization”,
Journal of Materials Processing Technology, vol. 155–156,
pp. 1132–1137, 2004.
N.P. Avdelidis, D.P. Almond, A. Dobbinson, B.C. Hawtin,
C. Ibarra-Castanedo, X. Maldague, “Aircraft composites
assessment by means of transient thermal NDT”, Progress in
Aerospace Sciences, vol. 40, pp. 143–162, 2004.
T. Inagaki, T. Ishii, T. Iwamoto, “On the NDT and E for the
diagnosis of defects using infrared thermography”, NDT&E
International, vol. 32, pp. 247–257, 1999.
G.M. Carlomagno, P.G. Berardi, “Unsteady Thermotopography in Non-Destructive Testing”, III Infrared
Information Exchange, pp. 33–40, St. Louis, Missouri, 1976.
G. Busse, A. Rosencwaig, “Subsurface imaging with
photoacoustics”, Appl. Phys. Lett., vol. 36, pp. 815–16, 1980.
G. Busse, D. Wu, W. Karpen, “Thermal wave imaging with
phase sensitive modulated thermography”, J. Appl. Phys.,
vol. 71, pp. 3962–5, 1992.
T. Sakagami, S. Kubo, “Applications of pulse heating
thermography and lock-in thermography to quantitative
nondestructive evaluations”, Infrared Physics & Technology,
vol. 43, pp. 211–218, 2002.
M. Choi, K. Kang, J. Park, W. Kim, K. Kim, “Quantitative
determination of a subsurface defect of reference specimen
by lock-in infrared thermography”, NDT&E International,
vol. 41, pp. 119–124, 2008.
D. Wu, “Lockin thermography for defect characterization in
veneered wood”, QIRT 94, pp 298–302, Sorrento, Italy,
1994.
W. Karpen, D. Wu, R. Steegmuller, G. Busse, “Depth
profiling of orientation in laminates with local lock-in
thermography”, QIRT 94, pp 281–286, Sorrento, Italy, 1994.
E. Grinzato, P. Bison, S. Marinetti, V. Vavilov,
“Nondestructive evaluation of delaminations in fresco plaster
using transient infrared termography”, Research in
Nondestructive Evaluation, vol. 5, pp. 257-274, 1994.
G.M. Carlomagno, C. Meola, “Comparison between
thermographic techniques for frescoes NDT”, NDT&E
International, vol. 35, pp. 559–565, 2002.
E. Grinzato, P. Bison, C. Bressan, A. Mazzoldi, “NDE of
frescoes by infrared thermography and lateral heating”,
QIRT 98, pp 64–70, Lodz, Poland, 1998.
X. Maldague, S. Marinetti, “Pulse phase infrared
thermography”, J. Appl. Phys., vol. 79, pp. 2694–8, 1996.
X. Maldague, F. Galmiche, A. Ziadi, “Advances in pulsed
phase thermography”, Infrared Physics & Technology, vol.
43, pp. 175–181, 2002.
W.J. Parker, R.J. Jenkins, C.P. Butler, G.L. Abbot, “Flash
method of determining thermal diffusivity”, Journal of
Applied Physics, vol.32, pp.1679–1684, 1961.
E. Grinzato, S. Marinetti, “Materials NDE by non linear
filtering applying heat transfer models”, NATO ASI, Series
E:Applied Sciences, Kluwer Academic Publishers, vol. 262,
pp.117–132, 1994.